Editing versus Encoding

[SCDVD]

I've noticed various comments in several discussion threads where encoders were described as "good" because they soften or sharpen video or in whatever way "improve" the video. It seems to me that a good encoder should faithfully encode the video with the fewest possible changes or "improvements". To me, the best encoder is the the one that outputs video that most nearly resembles the input video. The editor is where the filtering and "improving" the video should be done. If an encoder is sharpening or softening or making other changes to the video, you could say it is filtering the video. It seems to me that this is the wrong place to filter. It's a bit like putting a graphics equalizer on a component stereo power amp. Am I missing something?

[theewizard]

look at this way, you have a video, maybe for a customer or a friend, maybe something old you found on the web, you don't have the camera or source ( people involved ) you want to make this as viewable as possible, if its got noise/static in it OR mpeg artifacting from video capture, or maybe the camera focus was a little soft, or the lighting was wrong

being able to do something to improve things is important, this IS done at the encoding stage

editing video/film does not do this, editing selects what will be seen and what will be heard, it sets the pace, but editing tools do NOT effect the encoded video

its like asking the guy in the cutting room to change the lighting or camera focus of the film he is cutting, ( he can't do it ), this is what editing tools do , cut & join segmetns, and select-add a sound track NOT clean things up

[SCDVD]

theewizard - I understand what you are describing. When attempting to fix a video, you do what you have to do as best you can. I have done quite a bit of video fixing and each one is kind of its own case depending on what is wrong with it. I just fixed a DVD that had been produced by a non profit organization. They didn't have the source tapes anymore so I had to take it back to an .avi and used various filters in Vegas to "fix" it. I brightened, sharpened and dejittered it.

I am not describing the cutting functions in an editor, I'm talking about filters. When talking about an editor, generally both cutting AND filtering are involved. I then reencoded it and reauthored it. Sure, there is loss when reendoding but when you "fix" more than is lost on a bad video, you still have improved the video when you are finished.

The point I was trying to make in my original post wasn't so much to do with fixing videos but making them to start with. When you finish with the editing tasks of cutting AND filtering, you want an encoder to encode it without changing it. If you wanted it sharper or softer - or whatever else, you would have done it in your editor. Otherwise, to use my original example, you have a graphics equalizer on your power amp.

[theewizard]

well if you have editing tools that filter, you have tools i have never used

and if your worried about your encoding tool chainging what you filtered, then i guess your using the wrong encoding tool, because if you set it correctly it isn't going to change anything, changing things (when encoding ) are the options NOT the base set points

[edDV]

well if you have editing tools that filter, you have tools i have never used...

Wouldn't you call Virtualdubmod an editing tool that filters?
Also Premiere, Vegas et. al.

[thecoalman]

It seems to me that this is the wrong place to filter.

Is there a right or wrong place? As long as you can adjust the filters, turn them on or off etc I really don't see what the problem is. If your applying a filter in an external program or an encoder is applying one before it encodes relatively speaking it's not any different.

[guns1inger]

you want an encoder to encode it without changing it.

In an ideal world this would be true. And even in todays world it is, to a degree. Use an lossless encoder or encode uncompressed and you will get a copy of the video that is visually identical. You will also get huge files and it will only be playable on your PC.

When people talk about encoders softening or sharpening the image, they usually aren't talking about filters, but side effects of the way the encoder has been written. Some encoders do visibly alter the image in some ways under certain circumstances. All encoders will have some effect on the image if lossy compression is being applied. How much of an impact they have depends on the source, the encoder, and the amount of compression being used.

That said, many encoders do also have basic filtering tools built in (tmpgenc, ProCoder, CCE) that can be used if necessary. Sometimes a clip doesn't warrant a trip through an editor just to add a couple of degrees of brightness, so a filter is used at the encoding stage instead.

Finally, some people do like the effect that certain encoders have, or prefer one change to another simply as the lesser of two evils. I must say however that most of these discussions of this nature that I have read here revolve not around the positives of these changes, but rather as a negative feature of the encoder.

[mats.hogberg]

a good encoder should faithfully encode the video with the fewest possible changes or "improvements".

I couldn't agree more. The closer to the source mtrl the output becomes, the better the encoder. As simple as that. Period.

/Mats

[vhelp]

Last time was: April 29, 2007 Sun 8:58pm

The editor is where the filtering and "improving" the video should be done. If an encoder is sharpening or softening or making other changes to the video, you could say it is filtering the video. It seems to me that this is the wrong place to filter.

Well, actually, what you are experiencing with these encoders are there pre-filtering
process before the MPEG encoding. The video gets a (propreatory) filter, (ie, like in
Canopus, Procoder) where it pre-filters just before mpeg encoding.

So that is probably what you are experiencing.

Now, if you don't like what some of these mpeg encoders are doing to the source prior
to the mpeg encoding, then I would guess that your next step is to continue looking
for an mpeg encoder that does not pre-filter and only mpeg encodes a given source.

IMHO, the reason for each "branded" mpeg encoder's pre-filter should be an obvious
one.. in an effort to gain an advantage point. Now, sometimes, this is not evidensed
to some, as in your case -- you seem to notice this or some extent, there-of.

..another..

..Ok. Sometimes, it's not an pre-filter, but rather, it could be an (how shall I say it)
"artifact" of the mpeg encoder, as a whole. That is, as part of the process to making
the final mpeg, an artifact; or phenonima; or mistake; in terms of an aspect of the
process, somethere in there. Its probably not too far off in saying as an example
that an mpeg who is undergoing an upsize or downsize that this could be seen as a
clue for such artifacts, etc. This is an perfect example (and reason) why many suggest
that you not let the mpeg encoder do the up/down resizing, but to do it in post-process,
IOW, during the editing stage.

..another..

..is during a color space conversion. This can sometimes be seen as a pre-filter because
the final video looks soft, when really, its lighter looking. I've seen these in some of
my work, whenever I use AVIsynth and convert color space from RGB->YUV and back, in
error, on my part.

So, in conclusion, there are all sorts of ways to view this as pre-filter. But its a matter
of the user first identifying what they have; knowledge of their mpeg encoder (flaw, errors,
other phenomina) and how the user processes the video from Edit -to- MPEG.

a good encoder should faithfully encode the video with the fewest possible changes or "improvements".

Yes, this is true. But, until you identify such an mpeg encoder, you will have to keep
trial n error 'ing it until you do.

Here's some suggestions that might help:

RGB source: If you are going to use an RGB mpeg encoder.. ie, TMPGenc

Example: [RGB]VirtualDUB[RGB] -> [RGB]TMPGenc -> YV12 MPEG

Note, for RGB sources, it is wiser to stay inside this color space right up to the last
minute of the MPEG encoding stage, and let the MPEG encoder do the RGB->YV12 instead.

YUY2/YV12 source: If you are going to use an YUV (YV12) mpeg encoder.. ie, HCenc
** use an YUV (YV12) only editor (ie, AVIsynth) and frameserve to HCenc.

Example: [YV12]AVIsynth[YV12] -> [YV12]HCenc -> YV12 MPEG

So, if you are trying to process a video source to MPEG in hopes to duplicate as closely
as posible to the source, then it is assumed here, that you are not interested in including
any filtering. Then, the above examples would make sense.

-vhelp 4263

[edDV]

To me this is a consumer/hobbyist vs. pro issue.

The consumer hobbiest is on a linear track to make a DVD or some other singular task and doesn't report to a client. In that case, that person can do what he wants when he wants where he wants.

A pro project requires more discipline. Filters used during editing result in the director-producer's agreement on a final edit master. If it is a commercial or corporate project, the client sign's off at this point. Distribution to various formats follows. Encoder filters should only be used to achieve the best rendition of the master in that format. It is a craft decision rather than a creative decision.

Would you expect the TV or iPod version of Casino Royale to use a different color pallete than the DVD or BluRay version?

[SCDVD]

vhelp and edDV - Very good points. Another way to describe it is: Encoder filters are most appropriately used to normalize the output to match the master.

Many hobbyists use all-in-one software packages to make their own videos. With a few clicks of their mouse and keyboard they are able to accomplish the whole process from capture to burning. To them, there is no differentiation between editing, authoring or encoding. They are just "making a video" All considered, some of these software packages do a pretty decent job.